Journal of Dental Research, Vol 61, 824-827, Copyright © 1982 by International & American Associations for Dental Research Online Journals
Empirical verification of confidence intervals computed from dental data
M. E. Cohen and J. C. Cecil
Sampling experiments were conducted on five distributions of dental
epidemiological data which varied in degree of positive skew. The tendency
of nominal confidence intervals to be underestimated and asymmetric in the
two tails increased with skew and decreased with increasing sample size.
The findings conform to theoretical expectations and demonstrate the
importance of the normality assumption regarding sampling distributions. A
rule is presented which identifies a sample size so that the 95% confidence
probability statement would be incorrect no more than 6% of the time.
